A.S.

Williams DESIGN OF A LOW-COST OPEN-

SOURCE UNDERWATER GLIDER

Design of a Low-Cost Open-Source Underwater Glider

Alexander S. Williams

3/15/18

0
 A.S. Williams
Abstract
Typical buoyancy engine-based Underwater Gliders are
highly-complex and cost-prohibitive, generally ranging in
price-point from $50,000 to $250,000. A low-cost, Open-
Source Underwater Glider (OSUG) was thus developed as
a low-cost data-collection and research tool. This glider,
OSUG, is a sub-$1000, 1.2m long, 12kg, and capable of
50-hours of continuous operation. Its efficiency, and use-
case feasibility were evaluated. The buoyancy engine is
constructed of medical grade syringes that pull in water
from the environment to simplify the system and lower
costs. Direction of locomotion is controlled by altering
pitch and roll via changing the center-of-mass. The system
was designed to be primarily three-dimensionally (3D)
printed and fully-modular to limit cost and ensure
reproducibility.
Index Terms: open-source, underwater-glider, buoyancy
engine, underwater robotics, ocean environment
monitoring, mechanical design
Introduction
Underwater gliders are commonly used in oceanographic
research due to their capacity to collect data for long
durations and travel long distances1,2. Most underwater
gliders are based on buoyancy engines, that change their
buoyancy relative to the underwater environment by
changing their relative buoyancy to sequentially descend
and ascend in saw-tooth gliding patterns.
While some low-cost commercial gliders are available3,4,
including the Slocum Electric Glider Coastal made by
Teledyne Marine, these gliders still exceed $50,000. Some
low-cost academically-focused underwater gliders have
been created as well, such as Michigan Tech’s Research-
Oriented Underwater Glider for Hand-on Investigative
Engineering, ROUGHIE, which costs approximately
$10,000 in materials to reproduce. Unfortunately, current
underwater glider platforms are still reasonably high-cost
and do not publish the necessary documentation and files
necessary to reproduce them.
OSUG was developed to be the first fully-open source
glider as well as the lowest-cost underwater glider ever
developed, at less than 2% the cost of commercial
underwater gliders. Due to the dramatically-lower cost of
OSUG compared to a typical commercial system, numerous
OSUG systems can be deployed to collect data from large
regions as opposed to localized areas. Further, underwater
vehicles always have a risk of becoming lost on a mission,
1
DESIGN OF A LOW-COST OPEN-
SOURCE UNDERWATER GLIDER
especially when operated untethered5, or are not recollected
due to the high cost of locating and collecting the
underwater vehicle6. The sub-$1000 price-point of OSUG
makes recollection optional.
This paper is organized as follows: Section I discusses
OSUG’s mechanical architecture; Section II discusses the
electronics and control system, and Section III presents
plans for the continued development of OSUG.
Fig 1. OSUG Design and Prototype
Section I
Mechanical Design of OSUG
OSUG’s components are designed to be fully-modular, low-
cost, and replaceable. Thus, virtually all of OSUG’s
components are commercially-available or produced via
Fused Deposition Modeling (FDM) 3D printing. In doing
so, the barriers-to-entry of creating an underwater glider are
considerably lowered.
OSUG was designed in CAD software (Onshape) and is
fully variable-driven and parametric, such that a user can
easily manipulate numerous variables such as tube diameter,
number of syringes in the buoyancy engine, or leading-edge
nose taper angle. All parts were also created with a
predominantly-flat bottom-surface and few overhangs to
eliminate the need for support structures during 3D printing7.
To also account for the large deviations between different
low-cost FDM 3D printers, a set of calibration parts was
also created, such that users can print out the calibration
parts, measure them using standard metrology equipment
(low-cost calipers and micrometers), and input measured
values to automatically update the dimensions of OSUG to
ensure correct fit between components and subassemblies.
Fig 2. OSUG’s FDM 3D-Printed Components
 A.S. Williams
OSUG is comprised of seven subassemblies, including 1.)
the hull, 2.) external ballasts, 3.) a buoyancy engine, 4.)
hydrofoils 5.) the roll-control subassembly, 6.) the pitch-
control subassembly, and 7.) control system. All subsystems
are fully-modular to make manufacturing efficient and
replacement trivial. All other subassemblies connect to the
hull via leading and trailing-edge friction-fit end-caps.
The core of OSUG a hull made of a commercially-available,
pressure-resistant 4.5” outer-diameter 4” inner-diameter
acrylic tube, into which all modular assemblies are inserted.
To seal off the hull from the environment, two
commercially-available glider end caps from Blue
Robotics8 were utilized. These end caps are low-cost and
designed to have additional subassemblies attached to them.
When fully-assembled, OSUG’s hull is designed to be
mostly filled with only air, such that the OSUG has positive
buoyancy when submerged in a salt water environment.
OSUG has two 3D-printed ballast retaining rings, designed
to allow users to add or remove external metal ballasts until
OSUG is neutrally buoyant or slightly positively buoyant.
Finer adjustments can be made electronically by modifying
OSUG’s source code to change the amount of water in the
syringes required to achieve neutral buoyancy.
A tapered, 3D-printed nose cone, through which liquid can
be taken in and expelled for the buoyancy engine, mounts
to the leading-edge cap to seal the front-end of the hull. A
commercially-available plate with 3D-printed shroud seals
the trailing-end of the hull and enables application-specific
sensors to be added. Both end caps also serve to secure
OSUG’s major subassemblies. The buoyancy engine is
mounted to leading-edge end cap, while the roll-control
subassembly, pitch-control subassembly, and control
system are mounted to the trailing-edge end-cap.
Typical underwater gliders are based on buoyancy engines
using either 1) an oil-transfer system that pumps oil to and
from bladders, thus changing buoyancy relative to the
environment3,4, or 2) a thermal buoyancy system that melts
a waxy material that has greater volume when melted than
when solidified9, thus also changing the glider’s buoyancy
relative to the environment. OSUG uses a less common
form of buoyancy engine that modifies buoyancy by
drawing in and expelling water directly from the
environment. Generally, this is achieved via pumping water
into and out of an interior bladder in the glider.
However, OSUG utilizes a piston system to vastly simplify
the design. OSUG’s buoyancy engine consists of six 60mL
medical syringes, actuated via a NEMA 23 integrated-lead
2
DESIGN OF A LOW-COST OPEN-
SOURCE UNDERWATER GLIDER
screw stepper. Use of a stepper-driven piston-based system
allows the glider to take in and expel precise amounts of
water without complicated feedback systems associated
with a pump. When the stepper pulls the syringes’ plunger
shafts, a vacuum is created, and water is drawn in through
the leading-edge nose cone, thus making OSUG denser than
the surrounding water, causing descent; when the stepper
pushes the syringes’ plunger shafts, the glider expels water,
causing it to gain buoyancy and ascend. The buoyancy
engine has been cycle-tested to operate effectively to
approximately 250 hours, after which the syringes need to
be replaced.
Fig 3. Buoyancy Engine mounted to Leading-Edge End Cap
To course-correct against currents and external forces,
OSUG is also capable of modifying its center-of-mass by
controlling pitch and roll. To do so, OSUG’s batteries are
mounted to a pitch-control subassembly. These batteries are
actuated axially along the length of OSUG via a lead-screw
driven by a NEMA 17 stepper motor to shift OSUG’s center
of mass. While the buoyancy engine can be used to adjust
the glider’s pitch without a pitch-control subassembly, the
addition of a pitch-control subassembly allows a
considerably higher glide ratio (less aggressive slope of
travel) resulting in longer range and slower speed.
Alternatively, a lower glide ratio can be achieved, for more
rapid movement at the expense of range10.
Hydrofoils improve lift of the glider to increase glide ratio.
Each hydrofoil is made of 2mm-thick 316-alloy stainless
steel plate. A 3D-printed hydrofoil mounting ring holds
both hydrofoils approximately horizontal during gliding. A
fixed wing design would improve the lift-drag ratio11
however manufacturing of such a design would reduce
reproducibility.
To change roll and thus direction of travel, the entire pitch-
control subassembly is mounted to the roll-control
subassembly. The roll control subassembly consists
primarily of a NEMA 17 stepper motor with an attached
planetary gear set. The NEMA 17 motor is held stationary
and mounted to the trailing-edge end-cap, while the
 A.S. Williams
planetary gear set’s output shaft attaches to the rear of and
rotates the entire pitch-control subassembly. Thus, OSUG
can modify roll by rotating its center-of-mass and therefore
turn when gliding.
Fig 4. Pitch-Control, Roll-Control, Control-System
Subassemblies mounted to Trailing-Edge End Cap
The buoyancy engine needs to be attached to the back
assembly (consisting of the Pitch-Control, Roll-Control,
Control-System Subassemblies), both mechanically and
electrically. As the two main assemblies are friction fit into
the tubing, there is no lateral force transferred between the
assemblies. Three 8mm steel guide pins ensure that the two
assemblies retain rotational offset.
Section II
Electrical Design of OSUG
OSUG is controlled by a custom printed circuit board (PCB)
using an ATMEGA2560 microcontroller programmed over
the Fathom-S communication board using the full-duplex
RS-422 protocol. Standard USB is half-duplex, having a
timing requirement causing a cable length limitation of 5
meters12. The use of the Fathom-S communication board
allows for an extended tether length of over 600 meters.
Charging is controlled by a top-side lithium-ion charging
circuit, capable of 4A delivery. OSUG power is provided by
six energy-dense lithium-ion 18650 cells for a total capacity
of 85Wh. Each cell has an ultra-low-power integrated
voltage-, current-, and temperature-monitoring circuit for
battery protection as determined by the International
Electrotechnical Commission (IEC) 6213313 specification.
Charging and communication is achieved over a IP69K dry
mate underwater connector (Bulgin standard series), rated
to a 100-meter depth for prolonged periods.
The bottom-side Fathom-S communication board provides
5V supply and serial port communication to OSUG’s
control board via a 4-pin header. The Fathom-S onboard
power regulator provides low-noise 5V supply, with a
maximum current of 250mA, to the control board.
3
DESIGN OF A LOW-COST OPEN-
SOURCE UNDERWATER GLIDER
Previous designs used the ATMEGA328P microcontroller;
testing showed that more sophisticated control was required
and that the greater amount of flash memory of a
ATMEGA2560 was essential (256kB vs 32kB). A greater
number of Universal Asynchronous Receiver-Transmitter
(UART) ports allows for the expansion of the sensor array
to include Conductivity-Temperature-Depth (CTD)
sensors.
The control board uses a triple axis accelerometer and
gyroscope (InvenSense MPU-6050 sensor) for an inertial
navigation system to accurately determine OSUG’s position
underwater. The refresh rate was kept low (1.25Hz) and
reading acceleration only, in compliance with the
datasheet’s conditions for low power usage14.
A high-resolution pressure sensor (MS5837-30BA) was
used for depth sensing, as it has an internal oscillator for
long-term stability and gel protection with an antimagnetic
stainless-steel cap to ensure the module is water resistant.
The microcontroller interfaces with the pressure sensor over
an I2C bus interface to reliably transition gliding states at
specific depths.
Two Allegro A4988 stepper motor drivers are used to drive
the pitch and roll stepper motors. These are low current
applications, so the stepper motors deliver a maximum
current of 1A in these applications. The drivers are
connected via 8-pin headers as calibration stresses the driver
boards, sometimes requiring replacement.
The control board has an output header for SPI
communication protocol, driving an AMIS-30543 stepper
motor driver, mounted on the buoyancy engine
subassembly. The buoyancy engine stepper motor requires
high-torque, so this board provides up to 3A with forced air
flow cooling, varied via the SPI interface by the control
board from stall detection feedback.
Fig 5. AMIS-30543 Stepper motor driver mounted
adjoining the buoyancy engine NEMA 23
 A.S. Williams
The control board is also designed such that it can be used
as a slave board in conjunction with a Pixhawk 2.1 to
control OSUG. In addition to the open-source ArduPilot
UAV autopilot, the Pixhawk uses a high-reliability Global
Navigation Satellite System (GNSS) for autonomous
waypoint navigation.
The board was developed in the circuit board design
software, Autodesk EAGLE, as a two-layer board. Trace
widths were a default of 16mils for signal traces. The width
of traces used for power delivery were calculated using PCB
Toolkit V7.05, with a copper weight of 1oz and an allowed
temperature rise of 15OC. All components were Surface
Mount Device (SMD) variants, to reduce surface board-size
and cost. Through-hole headers were used for greater
structural strength.
Fig 6. Schematic and board view of OSUG’s control
board within EAGLE
Notably, the assemblies connected to the leading-edge end-
cap and the assemblies connected to the trailing-end edge
cap must be electronically-connected. To achieve this, there
are 14 Harwin 1mm through-PCB connectors, allowing for
a current limit of 10A per connector. The terminal pins are
4
DESIGN OF A LOW-COST OPEN-
SOURCE UNDERWATER GLIDER
13.2mm high, ensuring a good connection between the two
assemblies with a variety of separation distances that arise
due to dimension tolerances. Previous versions of this
connection system used pogo pins, however these provided
a force repelling the two assemblies, reducing reliability.
Pogo pins also only have a 3mm pin travel distance, which
was insufficient to deal with the variances in distance
between assemblies.
Fig 7. The buoyancy engine and back subassemblies
in unmated and mated positions
The ATMEGA2560 with Fathom-S communication board
allows the control board to be programmed via the Arduino
development suite, drawing upon standard libraries for the
high-resolution pressure sensor and stepper motor drivers.
Section III
Future developments of OSUG
The mechanical and electrical design of OSUG has been
outlined, additional software efforts are still in development
– especially with integration with the Pixhawk autopilot
suite. While the implementation is possible on current
 A.S. Williams DESIGN OF A LOW-COST OPEN-
SOURCE UNDERWATER GLIDER

hardware, further software development is required to
achieve full autonomy.
In the next version of OSUG, internal mechanisms will be
simplified to enable commercialization (while keeping the
OSUG open-source), increase reliability and cost optimize.
The current buoyancy engine system consists of six 60ml
medical syringes which results in limited durability of the
buoyancy engine. Currently, a new buoyancy engine with a
single custom-built piston is in development.
Currently OSUG can reliably descend and ascend
underwater, however future collaborations are to integrate
standardized sensor suites. CTD sensors are of importance
in underwater data monitoring missions for geophysical or
oceanographic research, for instance, to form
understandings of species and nutrient distribution or can be
used to determine other characteristics of underwater
conditions.
OSUG has reduced the high barrier of entry for long
distance underwater data collection set by commercial
systems. Continued development will enable far more
climate science to be conducted by both hobbyist and
private research groups. Given the open-source
development, as more developers adopt the platform there
will be a proliferation of varied capabilities.
[9] D. C. Webb, P. J. Simonetti and C. P. Jones, “SLOCUM: An
Underwater Glider Propelled by Environmental Energy”, IEEE
Journal of Oceanic Engineering, vol. 26, No. 4, 2001 pp. 447-
452
[10] N. Mahmoudian and C. Woolsey, “Underwater Glider Motion
Control”, 47th IEEE Conference on Decision and Control, 2008
[11] W. Zihao, L. Ye, W. Aobo and Wang Xiaobing, “Flying Wing
Underwater Glider: Design, Analysis, and Performance
Prediction”, 2015 International Conference on Control,
Automation and Robotics
[12] J Zyren and A Petrick, “Tutorial on Basic Link Budget
Analysis”, intersil, AN9804.1, 1998
[13] International Electrotechnical Commission, “Secondary cells and
batteries containing alkaline or other non-acid electrolytes –
Safety requirements for portable sealed secondary cells, and for
batteries made from them, for use in portable applications.
Lithium systems”, BS EN 62133-2:2017, 2017
[14] InvenSense “MPU-6000 and MPU-6050 Product Specification
Revision 3.4”, 2013
[15] A. L. Gordon and W. F. Haxby, “Agulhas Eddies Invade the
South Atlantic: Evidence From Geosat Altimeter and Shipboard
Conductivity-Temperature-Depth Survey” Journal of
Geophysical Research, vol. 95, No. 3, pp. 3117-3125, 1990
[16] K. Hayley, L. R. Bentley, M. Gharibi and M Nightingale, “Low
temperature dependence of electrical resistivity: Implications for
near surface geophysical monitoring” Geophysical Research
Letters, vol. 34, 2007

References

[1] C. C. Eriksen, T. J. Osse, R. D. Light, T. Wen, T. W. Lehman, P.

L. Sabin, J. W. Ballard and A. M. Chiodi, “Seaglider: A Long-
Range Autonomous Underwater Vehicle for Oceanographic
Research”, IEEE Journal of Oceanic Engineering, vol. 26, No. 4,
2001 pp. 424-436
[2] J Sjerman, R. E. Davis, W. B. Owens and J. Valdes, “The
Autonomous Underwater Glider ‘Spray’ ”, IEEE Journal of
Oceanic Engineering, vol. 26. No. 4, pp. 437-446, 2001
[3] G. A. Ribeiro, A. Pinar, E. Wilkening, S. Ziaeefard and N.
Mahmoudian, “A Multi-Level Motion Conroller for Low-Cost
Underwater Gliders”, Robotics and Automation (ICRA), 2015
IEEE International Conference on, 2015
[4] J. G. Graver and N. E. Leonard, “Underwater Glider Dynamics
and Control”, 12th International Symposium on Unmanned
Untethered Submersible Technology, 2001
[5] J. C. Kinsey, R. M. Eustice and L. L. Whitecomb, “A Survey of
Underwater Vehicle Navigation: Recent Advances and New
Challenges”, 2006
[6] R. Stokey, M. Purcell, N. Forrester, T. Austin, R. Goldsborough,
B. Allen and C. v. Alt, “A Docking System for REMUS, an
Autonomous Underwater Vehicle”, OCEANS '97. MTS/IEEE
Conference Proceedings, 1997
[7] S. Allen and D. Dutta, “On the Computation of Part Orientation
using Support Structures in Layered Manufacturing”, Solid
Freeform Fabrication Proceedings, pp. 259-268, 1994
[8] Available at https://www.bluerobotics.com, accessed 3/10/2018